Metrics

PPL	arc_easy	arc_challenge	piqa	winogrande	hellaswag	mmlu	QA Avg
16.73	37.84 ± 1.00	21.50 ± 1.20	61.43 ± 1.14	49.88 ± 1.41	33.52 ± 0.47	-	40.83

Training method based on BitDistiller Paper

Safetensors

Model size

1B params

Tensor type

BF16

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for BrownianNotion/TinyLlama_v1.1_2bit_int_7B_data

Base model

Finetuned

(53)

this model